Chess960 strategy: opening principles when there is no theory
In Chess960 your first move is made in a position you have never seen. Nothing you memorised applies. Everything underneath what you memorised does, and it turns out to be most of what you needed.
Why does opening theory not transfer?
Theory is a record of answers, and Chess960 changes the question. The Najdorf is a set of conclusions about one arrangement of pieces reached by one sequence of moves. Move the knight that starts on b1 over to a1 and every conclusion is void, including the ones nobody ever wrote down.
What survives is the reasoning that produced the theory. Develop toward the centre. Get the king off the middle file. Do not move the same piece twice while three pieces sit at home. Do not bring the queen out where minor pieces can gain time on her. Those are not simplifications for children. They are the compressed result of what strong players found, and the opening books are footnotes to them.
The practical difference is this. In standard chess, theory lets you play fifteen moves without consulting the principles. Chess960 takes the theory away and hands you back the principles. A club player with 1.e4 e5 2.Nf3 Nc6 3.Bc4 in his hands has three pieces out and a safe king by move six without thinking about it. Deal him bishops on a1 and h1 and he has to earn the same thing by counting.
Where does the corner rook go?
This is the problem that belongs to Chess960 and to no other version of chess.
RBBQNNKRIn standard chess both rooks are already solved. The h1 rook reaches f1 by castling, the a1 rook reaches d1 once the queen and bishop move, and nobody has to think about either. Shuffle the back rank and a rook can begin the game with a knight, a bishop and a queen standing between it and any file it wants. A rook on a1 with pieces on b1, c1 and d1 is not a badly placed piece. It is not a piece yet.
Castling is the cure, and it is often the fastest developing move available. The king and rook always finish on g1 and f1, or on c1 and d1, so one move can lift a rook out of the corner onto a central file and tuck the king away at the same time. In some arrangements the effect looks like a trick. Put the king on c1 with rooks on b1 and h1. Castling long leaves the king standing exactly where it already is and drops the b1 rook onto d1. One move: rook developed to the file your d-pawn is about to open, king safe, and the only price is the castling right you were going to spend anyway. If that reads as illegal, the castling rule is worth two minutes.
The same rule cuts the other way. Put the king on b1 with rooks on a1 and h1. Castling short walks the king from b1 to g1, and every square it crosses, c1 through f1, must be empty and unattacked. That means clearing four pieces off the back rank before you can castle at all. Castling long from the same position moves the king one square and puts the a1 rook on d1.
So before you play anything, count the squares each way. The side that needs less clearing is usually the side you castle, and that decision tells you which pawns to push and where the minor pieces go. It also decides the rooks. Whichever rook castling does not solve is the piece you will still be shuffling on move twenty, and the pawn you push on that wing is the one that eventually frees it.
Which pawn is defended only by the king?
The most attacked square in beginners' chess is f7, and the reason is an accident of the standard array. Look along White's second rank in the normal position. The a2 pawn has the rook behind it. The b2 pawn has the c1 bishop. Both c2 and d2 have the queen. The e2 pawn has the queen, a bishop, a knight and the king. The g2 pawn has the f1 bishop, and h2 has the rook. The f2 pawn has the king, and nothing else, because the bishop on f1 stands on a light square and f2 is dark. Scholar's Mate exists because of that one gap.
RNBQKBNRChess960 keeps the gap and moves it somewhere else. In most arrangements one or two pawns on the second rank are defended by the king alone, and they are usually not on the f-file. Find yours before you touch a piece.
Then find your opponent's, which is on the same file, because Black mirrors White exactly. This is the part most explanations leave out and it changes how you should play. The weak square is symmetrical. It is not a flaw in the position handed to you, it is a target both players can see, and the opening becomes a race to get a piece pointed at it first. Often the whole first phase of the game is decided by whether the defender spends a move closing the account, with a pawn push, a rook lift, or a knight, and by whether that move cost him the initiative he needed.
What is a bishop doing on a1?
A bishop that starts on a1 or h1 is already sitting on the long diagonal. It does not need developing, it needs unblocking. In standard chess a fianchetto costs two moves, g3 and Bg2. When the bishop starts on h1 the pawn move is the entire plan, and the bishop never leaves the corner, which is also the one square where your own g and h pawns can shut it in for the rest of the game if you push them carelessly.
BBQNNRKRBishops on the b1 and g1 squares are the ones that surprise people. A bishop on b1 stands on the diagonal that ends at h7. Your own c-pawn blocks it, and your c-pawn is one move from not blocking it. A bishop on g1 stands on the diagonal ending at a7, behind the f2 pawn. In both cases nobody has moved yet and the gun is already aimed at a square next to where a king may well castle. Check whether your opponent has the same bishop, because he does, on the mirror square.
What should you look at before move one?
Take the thirty seconds. In this order:
- Your king, and the two castling counts. How many squares must clear for each side, and which of them can happen inside four moves.
- The pawn on your second rank that only your king defends, and the same file on his side.
- Your bishops. One is usually looking at something already; the other is usually behind three pieces. The buried one decides which pawn you push first.
- Your knights. A knight in the corner is the slowest piece on the board. A knight starting on a1 needs four moves to reach c3, not two. Trace it and you will find no two-move route exists, and the square colours forbid an odd number, so four is the answer.
- The rook that castling will not solve, and the plan that eventually frees it.
What are the first ten moves for?
Making the position look like something you have seen. You are not playing an opening. You are steering toward a middlegame you already know how to handle.
The pawns are the steering wheel, and here is what transfers whole: every Chess960 game begins with all sixteen pawns on the squares they occupy in standard chess. Every structure in every book you own is reachable from every one of the 960 arrangements. An isolated queen's pawn is still an isolated queen's pawn. A closed centre with pawns locked on e4 and d5 behaves as it always did. Your entire experience of pawn structure is valid from move one, and it is probably the largest single asset you bring to the board.
So push the centre pawns, put a knight on a square whose name you know, castle early, and by move eight you have a position that is unfamiliar in its piece placement and completely familiar in its skeleton. Resist the urge to punish the arrangement. Most positions contain no early tactic, and hunting for one costs you the two or three tempi that would have bought a normal game.
Do strong players prepare for Chess960?
Yes, and it is worth saying so plainly, because the claim that theory is impossible here is false and was never the point.
Each of the 960 arrangements has a number, and engines have been pointed at them one at a time. Analysis indexed by starting position exists, some of it published, more of it private. When an event announces the position before the round, both players open a laptop and do exactly what they would do with a Sicilian. At that moment Chess960 is a memory contest with a smaller library.
What differs is depth and shelf life. Standard theory is a hundred and fifty years of work on one position, tested in millions of games. Chess960 preparation is a few hours of engine time on an arrangement your opponent will probably never meet again. It is real, and it is thin. For a club player it is not a factor at all, because nobody has prepared the position you are about to be dealt.
How do you practise this?
Deal yourself twenty positions with the position generator and do nothing but the checklist: king, weak pawn, bishops, knights, the awkward rook. Thirty seconds each, no moves. It takes ten minutes and it is the fastest way to stop feeling lost on move one. Then play the computer from a random array, and use the puzzle trainer for the part principles cannot help with, which is seeing a fork on b1 when your eye has only ever filed it on f7. If you want a thousand of a thousand of those, the books hold that many each. For what the two games do and do not share, there is a separate piece on Chess960 against standard chess.